One of the most challenging tasks in data backup is the capture of files that are open or in use by an application. Many users believe that if they select a file for backup, for example the Outlook PST file that contains your email, contacts and calendar, it will be included whenever a backup occurs. Unfortunately, they often find out when attempting to recover the file that it has not actually been backed up.
Normally, Operating Systems like Windows do not permit users to perform any action on an open file and since applications like Outlook are rarely shut down by most users, the open file could not be backed up. So how do you ensure that open files like your Outlook PST are backed up in order to ensure you can perform data recovery in case of a disaster?
Open files can be backed up using several different methods, including: Locked File Backup (LFB) and Application Program Interface (API).
Locked File Backup (LFB)
The LFB method is a generic approach that backs up all files whether they are open or closed. It can be used as a stand-alone or an integrated process and is often used with relational databases like Microsoft SQL and Exchange.
LBF is normally used if there is no API technology available. The Application Program Interface method uses programming routines built into the cloud backup software to guarantee file integrity as well as other options that are not available in the LFB method.
The API can be used for specific databases and contains programmable calls that can be used by other applications to read and write to the database. Most cloud backup service providers that offer backup of open files use both methods to ensure that files in use by an application are captured.
When LFB is used, a static image of the open file is maintained on one partition of the disk in readiness for a verification pass (where a block of data on the backup media is compared with the block of data on the disk) and backup. If the LFB program -- that has been activated on open files -- signals that it is okay to proceed and that all related files have been synchronized, the backup commences.
A dynamic pre-write cache is automatically allocated and a standard disk file is created for all open files within the partition that is being processed by the backup program. All write operations are then directed to the correct file as the pre-write data is placed in the pre-write cache. If the file gets modified during the backup process, the LFB software uses the pre-write cache to satisfy read requests. The pre-write cache is released after the file has been written. This process also does not interrupt other tasks that are being performed on the system.
When files are open and are being shared by multiple users on the network, backup is done using Open Shared LFB. At the start of the process, the open file is written into a backup report by locking the file, backing it up and then releasing it. The changes that happen subsequently, are stored in paged memory and get re-synchronized with the backed up locked file. For relational databases that are open in the shared mode, an integrated locked backup method has to be used.
In applications such as Excel, Lotus Notes and others, open files are locked against other agents. It is still possible for a LFB program to take advantage of a pause in the application to capture and freeze the file and start the backup of those files. This ensures that the files will be in synchronization at all times during backup. All changes in the file subsequent to the freeze are placed in the pre-cache for verification and synchronization. The drawback of this method is that it cannot determine the accuracy or integrity of a file and is not used with large databases that have millions of records that need to be backed up.
In Part II, we will discuss more about Application Program Interface (API) method to backup open files.